It is actually pretty stable, least at 200-600 rpm, much smaller rpm is a bit more problem, because eye does not "make" the dot track continuous. I first put the drawbolt only on finger tight and because this unit is not balanced it would shake it loose and I could see it it on the laser beam. Hmm...wonder how small dynamic runout (imbalance) it could show........

It might make better dynamic balancer fixture

I'm just trying prove here the concept and figure out what it is for me.
My first attempt is flawed, because I want to dispense with the arbor. It should be mounted over the bottom part of the spindle or such to allow drill/mill bit to coexist. It also has sharp edges and design is dangerous. I was thinking of using a holes saw to make a disc out of plywood and drill few holes to different angles to mount the laser. But it would have been slower to make.
But sheet metal is easy to fabricate, easy to bend to try out different angles and such.
Now I have to try it out with whole lot more ambient light to proved myself will that simple askew mounted 5mW laser work for me or not. Lucily this 5 mW laser is probably way under nominal value, but it's not the problem here.
If it doesn't then I have to look for 1mW unit or start thinking of attenuating it with ND filter or polarizing filter. If I choose filter approach, then I'm very tempted to fold the optical axis with prism or two. This would have a whole lot more involved structure with mounting the prisms and other optics, but it would be easier whole lot more versatile.
I took the collimating lens of the laser and the exit beam from the laser diode was really really wide and oval. Collimating lens has really big impact on dot quality. Not sure how deep I would like to get on this one.
Pekka